The problem is to find numerical methods which will allow three dimensional reconstruction of the breast so as to: 1) obtain 0.1 mm spatial resolution; 2) use relatively few mammograms taken at different angles; 3) be computationally feasible; 4) be implementable as an inexpensive xray breast scanner. A method for dividing the space representing a computed cross- section has been found which is 10 to 20 times faster than existing methods and allows the finest details to be reconstructed.